May 1998 - Gpc

Delphis Object Pascal (was: RE: Reformatted: EFLIB as class library for Object-Pascal (fwd))
by Andreas Prucha [helicon] 08 May '98

08 May '98

Hi, >- Properties: I have them listed in my To-Do-list > (http://fjf.gnu.de/gpc-todo_toc.html) but I personally don't know > too much about them. Are they a Delphi extension? If so, are there > any documents available to describe them, or otherwise, could some > Delphi user put together a list of features required that could be > used as a "standard" that GPC, and perhaps FPK, can implement? Well, they look like fields, feel like fields, but aren't fields. Properties are declared: Property <name> : <type> read <howtoget> write <howtoset>; Proprties can read/write directly to a field or use get/set methods: Type TMyClass = Class (TObject) private FValue : Integer; public property Value : Integer Read FValue Write FValue; end; using the object it can easily be accessed like a field: MyObject.Value := 256; x := MyObject.Value; Data is still encapsuated. It's easy to change the implementation. Let's assume that Value may not be > 255. Type TMyClass = Class (TObject) private FValue : Integer; Procedure SetValue (Value: Integer); public property Value : Integer Read FValue Write SetValue; end; Procedure TMyClass.SetValue (Value: Integer); begin If Value > 255 then Raise Exception.Create ('Value may not be > 255'); FValue := Value; end; If the user now does MyObject.Value := 256; the assignment will fail with an exception. >- Classes: At http://pascal-central.com/OOE-stds.html, there is a > draft for an ANSI Object Pascal standard. AFAIH, Delphi's > implementation is quite close to this standard. One of the plans > for GPC is to implement this standard draft without dropping the > BP object model that is already implemented, in the hope that > there won't be great conflicts between the two. They are quite close. The look the same at the first glance, but they are not compatible. For example, Delphi does not know Class-/Kinds/. Classes are always declared the same way: TNewClass = Class (TDerivedFrom); Encapsulation is done quite the same way as in BP: TNewClass = Class (TObject) private // May not be accessed from outside (also not in subclasses) protected // May not be accessed from outside, but subclasses cann access public // May be accessed from outside published // A IDE-thing. should be handled the same way as "public" end; Othere than the Ansi-Draft in Delphi not each method is virtual. Delphi supports "Virtual" and "Dynamic". Procedure x; Virtual; Procedure y; Dynamic; "override" is used to override the method; Procedure x; override; Create is the standard-constructor and destroy the standard-destructor. Objects are always on heap and the pointer is automatically dereferenced. Objects are initialized the following way: MyObject := TMyClass.Create; and freed with MyObject.Free; Since they are automatically dereferenced it is not necessary to say MyObject^.Free; One of Delphis's bith strengths are the runtime-type-informations. Delphi introduces "is" and "as". "is" checks wheter the object is this or an subclass: If TMyObject is TMyClass then do ...; If TMyObject is a object with type TMyClass or a object of a subclass of TMyClass then the expression is true. "as" is used for typeconversion: try x := (MyObject as TMyClass).Value; except x := 0; end; If the typeconversion to MyClass is not possible (not (MyObject is TMyClass)) as raises an exception. There is much more to say about Delphis Class-Model, but these are the basics. In Delphi both models live together. The old model can be still used, but it is not possible to use the new model to subclass from an old-model-type object. But this is not necessary since in Delphi each class uses the new model. >- Exception handling: I myself would like to use it (for OOP as well > as non-OOP programs), so I'd be willing to put some effort into > implementing it when I have the time (later this year, hopefully). > However, as with properties, I'm lacking a (pseudo-)standard. > I do know the Java exception handling, and I think it's not too bad > and could be implemented similarly in Pascal. OTOH, I read that the > latest GCC version (on which GPC is based) has introduced exception > handling, so it might be easier to implement this if we can reuse Exception handling - really nice thing in Delphi. An exception is fired with raise. The information about the exception is stored in an object. The basic-exception-class is "Exception" raise Exception.Create ('This was wrong'); To use store additional data (e.g. and errorcode) exception can be easily subclassed: EMyException = Class (exception) private FErrorCode: Integer; public Constructor Create (Errorcode: Integer); property Errorcode : Integer Read FErrorcode; end; Constructor EMyException.Create (Errorcode: Integer); begin FErrorCode := Errorcode; Inherited Create ('Error '+IntToStr(Errorcode)); end; Exceptions are catched the following way: try // code U try to execute except // handle the exception end; Since exceptions are classes the different types of exceptions can be handled in "except" try // code U try to execute except On EMyException do xxxx; // Do special exception handling here On EAnotherException do xxxx; // What you want else raise; // let the standard exception handler do the thing end; "finally" is used to assure that some piece of code is always executed: try // code U want to execute finally // Do it also if code fails end; This is very usable to assure that resources are freed: MyObject := TMyClass.Create; try x := MyObject.value / 0; // Ooops /0 exception here! finally MyObject.Free; // But the object will be freed. end; > that code. OTTH, if there already is an implementation in a Pascal > dialect (Delphi?), and it's suitable, it would be a good idea to be > compatible to this. So again, could someone make a list of required > features (perhaps with some sample code)? I totally agree. IMO GCC should be compatible with Delphis Object Pascal, since OP is the real standard (1000000+ sold copies of Delphi) and not only a paper-standard. But you guys do it for free, so everything is fine, but Delphi/OP compatiblity would be /great/. If people come to Linux they could easily reuse existing Delphi/OP-code. And yes. Delphi's implementation is /great/ even the new referece counted string types make live much easier. They can save more than > 255, are dynamically allocated and referece-counted. And can for type-conversion easyly typecased with PChar: Procedure MyProc (xxx: PChar); var mystring: String; MyProc (PChar(MyString)); This type of string was introduced with Delphi 2. There are much more things in Delphi/OP which make live easier. The big advantages I see at the moment: - New class model with RTTI, Properties - Exception Handling - AnsiString and WideString (8bit and 16bit-char strings with > 255 chars, String is an AnsiString) - - Interfaces (not really necessary but nice) You asked if there is an description of Delphi/OP. I think there is an helpfile available on www.interprise.com (I'll look if I can find it). But I doubt that it is possible to implement it compatible w/o having a look at Delphi itself. But I hope that I could explain some basic things. And thank you guys for supporting (Object) Pascal. Andreas

2 1

None
by hantschr＠hantsch.co.at 08 May '98

08 May '98

--note----------------------------------------------------------------- Username: [rainer] Use the above line in your e-Mail body to reach me in our company. You may use [rainer] at the beginning of the SUBJECT line (with higher priority) instead. -- This will allow us "Automatic Mail Forwarding". Our human "postmaster" may be reached with [rainer]. DO NOT USE OTHER FILE-ENCODING THAN UUENCODE, IT CANNOT BE PROCESSED!!! --note----------------------------------------------------------------- Received: from Hantsch-Message_Server by hantsch.co.at with WordPerfect_Office; Fri, 08 May 1998 08:55:30 +0100 Message-Id: <s552c882.001(a)hantsch.co.at> X-Mailer: WordPerfect Office 4.0 Date: Fri, 08 May 1998 08:54:16 +0100 From: hantschr(a)hantsch.co.at To: gpc(a)hut.fi Subject: GPC for LINUX Item Type: Note Hello, there! Some months ago I tried GPC, which came with my SuSE 5.1 distribution, but I could not use it. The reason: ----------- GPC DIDN'T ALLOW THE SIMPLEST TURBOPASCAL COMMANDS, f.e. assign(). This and the missing documentation made it nearely impossible to write programs which do more than printing "Hello world..." to the Screen. So I stopped with it immediately because I didn't want to waste time. __________________ Now I definitely have to develop some smaller applications/tools for my Linux server (now SuSE Linux 5.2) and so I am again in the same situation of searching for a *WORKING* compiler. I have heared from a compiler which is VERY similar to TurboPascal and also includes an IDE. I guess, this is NOT GPC - am I right? Anyway, I hope that GPC has also changed very much and become usable. Has it meanwhile become much more compatible to Turbo-/Borland Pascal? I would like to be able to compile DOS programs with a minimum of changes with GPC. Please be so kind and leave me some information. Thanks in advance, Rainer Hantsch (hantschr(a)hantsch.co.at) ---------------------------------------------------------------------- | Office address: | You may reach us by: | |--------------------------|-------------------------------------------| | ING. RAINER HANTSCH | Phone : +43 1 7988538 0 | | Khunngasse 21/1 | Fax : +43 1 7988538 18 | | A-1030 Wien/Vienna | BBS : +43 1 7988538 17 | | | e-Mail: hantschr(a)hantsch.co.at | | *** A U S T R I A *** | www : http://members.eunet.at/hantschr/ | ----------------------------------------------------------------------

2 1

Re: Another more striking example
by Kevin A. Foss 08 May '98

08 May '98

On Thu, 07 May 1998 23:12:03 +0200, FPL wrote: >program loop2(input, output); >var > i, j : integer; > >begin > j:=0; > repeat > for i:=0 to 10 do > if (i div 2)=0 then > begin > write('*'); > j:=j+1; > end; > writeln; > until j=10; >end. > >Do you agree the output should be a rectangular array of 5 stars by ten, >namely, something like No. Note that for i from 0 to 10 "i div 2" will only equal 0 when i = 0 and when i = 1, otherwise, 2 will go into i at least once. This is what I get for output: ** ** ** ** ** Which is correct, as you increment j everytime you print a star. Putting in a write(j); before you increment j may help you understand what is going on. If you want 10 rows then you need to increment j outside of the 'if' block, which is probably most easily done by just removing the begin/end pair. [...] if (i div 2) then write('*'); j := j + 1 writeln; [...] If you want 5 stars across, then make i go from 1 to 10 and change the test to (i mod 2). hope this helps, -Kevin -- Kevin A. Foss --- kfoss(a)mint.net

4 3

Re: Reformatted: EFLIB as class library for Object-Pascal (fwd)
by Frank Heckenbach 08 May '98

08 May '98

Johan Larsson wrote: > I am the author of EFLIB, a free object-oriented extension for > Object-Pascal which currently is evolving but will be released in a brand > new version soon. Our software gets more and more mature for every day. We > now want to make EFLIB a standard template library for both of your > languages: GNU Pascal and FPK Pascal, and we want to support CORBA, X11, > Windows NT and even Motif. Also, we want to keep the project close to the > state of art in computer science. We want to incorporate only modern > algorithms, modern solutions and - most importantly - only modern > approaches to software reuse: the design patterns. > > EFLIB is free software (http://www.csd.uu.se/projects/EFLIB/Support/FAQ/) > We believe that a free mind "is-a" creative mind. However, EFLIB will also > be wrapped and sold in a commercial CDROM kit. We believe that free > software can indeed be commercial software as well (as do Richard Stallman > for that matter). To warrant the freedom of distribution and use, we use a > GNU LGPL-compatible license agreement (currently being completed by a > Swedish lawyer). I appreciate that it will be free software. However, due to the recent creation of the NPL, MPL and other more or less compatible free licenses, I'm a bit wary that it will really be a GPC compatible license. I looked at the license mentioned above, but I'm not a lawyer :-), therefore I'd like to see an official statement confirming the possibility to combine it with GPL/LGPL licensed software (i.e. to use and distribute without conflicting with either of the licenses). I hope this will be possible. > We think EFLIB will prove to be a great value for the future of > Object-Pascal. We believe that the next step of programming will be an > increased level of abstraction due to the introduction of design patterns > and other higher-level entities of software design. We also believe in > modelling and tool support for software design. We want to combine all of > these things in EFLIB. We are also considering integrating modern > technologies such as CORBA and Motif support. The trend goes to language > and platform independence and universal modelling and reuse, and that's > exactly what EFLIB is all about. GPC is also about platform idependence and easy to integrate with other languages, so this could be a good combination. :-) > I am, as the project leader of EFLIB, confident that EFLIB will fit actual > needs as an official standardized class library for FPK Pascal and GNU > Pascal, and partially for Borland Delphi 3. However, to reach this goal I > will need collaboration from all of you. Many people have already reacted > positively to EFLIB in the FPK and GNU domain. First of all, we need > developers who can help us with the actual coding and testing, and give us > response to the specifications. We need experienced CORBA / ORB / DCOM > developers. We need experts on design patterns and reuse, and experienced > developers in X11, Motif and Windows NT. We also need some changes in the > GNU and FPK Pascal standards: full support for properties, class model and > exception handling. It could well be time for a precise standard to settle > between GNU and FPK. These things are actually planned for GPC: - Properties: I have them listed in my To-Do-list (http://fjf.gnu.de/gpc-todo_toc.html) but I personally don't know too much about them. Are they a Delphi extension? If so, are there any documents available to describe them, or otherwise, could some Delphi user put together a list of features required that could be used as a "standard" that GPC, and perhaps FPK, can implement? - Classes: At http://pascal-central.com/OOE-stds.html, there is a draft for an ANSI Object Pascal standard. AFAIH, Delphi's implementation is quite close to this standard. One of the plans for GPC is to implement this standard draft without dropping the BP object model that is already implemented, in the hope that there won't be great conflicts between the two. - Exception handling: I myself would like to use it (for OOP as well as non-OOP programs), so I'd be willing to put some effort into implementing it when I have the time (later this year, hopefully). However, as with properties, I'm lacking a (pseudo-)standard. I do know the Java exception handling, and I think it's not too bad and could be implemented similarly in Pascal. OTOH, I read that the latest GCC version (on which GPC is based) has introduced exception handling, so it might be easier to implement this if we can reuse that code. OTTH, if there already is an implementation in a Pascal dialect (Delphi?), and it's suitable, it would be a good idea to be compatible to this. So again, could someone make a list of required features (perhaps with some sample code)? -- Frank Heckenbach, frank(a)pascal.gnu.de http://fjf.gnu.de/ PGP keys: http://pgp5.ai.mit.edu/pks/lookup?op=index&search=Frank+Heckenbach

4 3

Another more striking example
by FPL 07 May '98

07 May '98

Peter, list, I think this program is a better example of what the problem is. It is not as artificial as the prior one, and can be executed without gdb so that one can see the problem may lie somewhere else that in debugging informations. program loop2(input, output); var i, j : integer; begin j:=0; repeat for i:=0 to 10 do if (i div 2)=0 then begin write('*'); j:=j+1; end; writeln; until j=10; end. Do you agree the output should be a rectangular array of 5 stars by ten, namely, something like ****** ****** ****** ****** ****** ****** ****** ****** ****** ****** ? Well, what *I* get when running the output code is * * * * * * * * * * Is it the same with you? (You just cannot imagine how scared I am that *my* and only mine GPC version, poorly installed, is bugging this way). Also, I have (in the other program) some file of text type ever opened, and I have noticed in applying your advice that GPC complains it cannot write on the default output device (the screen) because of my opened file. Just a writeln('*') and *not* a writeln(f, '*') wille make it say ?GPC runtime error: File is not open for writing (#9) Cheers. F.P.L

1 0

Is it a bug?
by FPL 07 May '98

07 May '98

I have a piece of code that does not follow my understanding of pascal. As I don't want to bother you in reading long code, here is a tiny model, that (mis)behave the same. 1=1 and 1=2 are to avoid discussing "when it does occur" or "once it stopped". I just don't know wether it is a bug or an error of mine. 1 program model(input, output); 2 var 3 i : integer; 4 5 begin 6 repeat 7 for i:=1 to 10 do 8 if 1=1 then 9 writeln('*'); 10 until 1=2; 11 end. As I understand it, once reached line 7 (the 'for' loop), the following instruction ('if then') is executed ten times, so the output consists in a row of ten stars as the 1=1 is always true. Then the 'repeat until' loop is executed, forever since 1 never equals 2. It is not obvious the programme does not work if one executes it, but the debugger will betray another behaviour (if you change something in the structure, it is not certain it will still "not work"). Once reached the 'for' statement, only one star is printed then the pointer directely jumps on repeat, not printing ten stars in a row, just one. Nothing is changed in 'begin end;' insertion in the code. If it is some kind of bug, I theoretically use last GPC beta version. Cheers. F.P.L

2 3

Re: How it is compilated?
by richard.kerry＠quantel.com 07 May '98

07 May '98

It has to call(a) twice. There may be something else being done in call(), globals read/altered, files used, hardware accessed etc. VAX Pascal has a directive, called something like 'deterministic', which means that a routine, when called with the same parameters, will return the same result and not affect anything else. The alternatives are a scheme like this, or the compiler analyzes the routine and determines whether it is deterministic or has any side effects. I'm sure our compiler writing friends would be happy to spend a very large amount of time writing this. This is related to the reason why short-circuit evaluation should not be done unless requested. Some other bit of program may rely on all the components being evaluated if they are functions. Regards. Richard.

1 0

How it is compilated?
by FPL 07 May '98

07 May '98

Hi Peter, Hugh Chief, Howdy the list. I think one good reason why peoples interested in programming should know a bit about compilers is that they hence have an idea how to write their code so that the compilated version of it won't be too bad. I know damn shit about compilers, but there is this situation I am always perplex about when facing it. Suppose "call" is a big complicated function, say, a theta(2**n) one, really the big one. How will GPC translate the following line into machine code? if call(a)=1 or call(a)=2 then ... Will it call two times the function? Or will it by itself do the simplification? Should we by ourselves complicate our way of writting code in defining another variable to store in advance the result of call(a)? Cheers (and thanks Peter for your proposal of implementing set of records, I use something near it, it works fine). F.P.L

2 1

set of record
by FPL 06 May '98

06 May '98

Hi the list, This appear not to be a GPC relevant question, sorry. Cannot we define set of record? type cellule = record x, y : integer; age : integer; end; population = set of cellule; This does not compilate, is it a language non sens?

2 1

Initialisation while declaration
by FPL 05 May '98

05 May '98

Hi the list, Does GPC support a feature kindwise those of C compiler that is to define the content of something while you declare it? I mean, assume I want to define one integer then to set it to naught, C allow me to replace int c; c=0 with int c=0; Is GPC can do the same than elder brother gcc? If it can, is it a GPC feature or is it Borland compatible? And how to do it to initialise each field of a record? See ya! Fabrice Llaussy "Rune's Rule: If you don't care where you are, you ain't lost." "Look out fo' flyeeng feet" O__/ \ laussy(a)nat.fr \/|_____. - O <__.\/\> / / \ __________|___\______ http://ourworld.compuserve.com/homepages/BigApe323

5 4